ascend-moe-optimizer-trace-analyzer

在用户提供 Chrome/Perfetto trace.json、或排查 Ascend 上 MoE/FusedDeepMoe 等算子性能时使用。按 phase、category、core group、tid 统计耗时、overlap、bubble，输出 CSV、Markdown 报告与确定性诊断；可选外部 LLM 扩写分析。默认 phase 映射面向 UMDK FusedDeepMoe，其它 trace 需替换或扩展 config/phase_map.yaml。

Quality

78%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./jiuwenclaw/jiuwenclaw/resources/agent/jiuwenclaw_workspace/skills/ascend-moe-optimizer-trace-analyzer/SKILL.md

Quality

Content

63%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A thorough, actionable body with concrete commands and a clear execution sequence, weakened primarily by length and the total absence of progressive disclosure — substantial reference material is inlined rather than split into referenced files.

Suggestions

Move the detailed output-file catalog, '指标口径' definitions, and core-group/phase-category rules into separate reference files (e.g. OUTPUTS.md, METRICS.md) referenced one level deep from SKILL.md to improve progressive disclosure.

Tighten or remove the '能力概览' section, which restates the frontmatter description, and condense the parameter list by merging the basic and default-phase-map command examples.

Add one fully concrete example invocation with realistic sample values (not just placeholders) so a common case is copy-paste ready.

Dimension	Reasoning	Score
Conciseness	Content is dense and domain-specific rather than padded with general knowledge, but at ~210 lines sections like '能力概览' overlap with the description and the output-file catalog and '指标口径' could be tightened; mostly efficient with room to trim.	3 / 5
Actionability	Provides concrete executable commands ('python3 app.py --trace <TRACE_JSON> --phase-map <PHASE_MAP> --output-dir <OUTPUT_DIR>') with a documented parameter list and multiple variants, but examples rely on placeholders rather than a fully copy-paste-ready concrete run.	4 / 5
Workflow Clarity	Clear sequence: confirm context (TRACE_JSON/OUTPUT_DIR/PHASE_MAP/etc.), choose phase map, cd to skill dir, run, then optional LLM analysis, with a '基础验证' command; minor validation gaps but the operation is read-only analysis so the destructive-cap does not apply.	4 / 5
Progressive Disclosure	Well-organized into labeled sections, but no bundle files exist and all reference-like material (output file catalog, metric definitions, core-group rules, phase/category config detail) is inlined in one ~210-line SKILL.md with no one-level-deep references to split it out.	3 / 5
	Total	14 / 20 Passed

Description

92%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A highly specific, well-structured description that clearly states both capability and trigger conditions with concrete domain terms. Its only minor weakness is that trigger keywords are densely technical, which is appropriate for the audience but leaves a few natural-language variations uncovered.

Dimension	Reasoning	Score
Specificity	Names multiple concrete actions — '按 phase、category、core group、tid 统计耗时、overlap、bubble' and '输出 CSV、Markdown 报告与确定性诊断' plus optional '外部 LLM 扩写分析' — giving comprehensive coverage rather than a single generic verb.	5 / 5
Completeness	Explicitly answers both: the 'what' (aggregate timing/overlap/bubble, output CSV/Markdown/deterministic diagnosis) and the 'when' ('在用户提供 Chrome/Perfetto trace.json、或排查 Ascend 上 MoE/FusedDeepMoe 等算子性能时使用'), with concrete trigger phrases.	5 / 5
Trigger Term Quality	Strong natural/keyword coverage including file extension form ('Chrome/Perfetto trace.json'), domain synonyms ('MoE/FusedDeepMoe', 'Ascend'), and trigger phrasing ('排查…算子性能'), but the terms are heavily technical jargon with a few layperson variations absent.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (Ascend MoE/FusedDeepMoe trace performance analysis) with distinct, specific triggers and minimal realistic overlap with other skills.	5 / 5
	Total	19 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: XiaoLuoLYG/GOD
Path: jiuwenclaw/jiuwenclaw/resources/agent/jiuwenclaw_workspace/skills/ascend-moe-optimizer-trace-analyzer/SKILL.md
Commit: da0d716

Reviewed: 2 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.